R6RS

Scheme, the dialect of Lisp invented by Guy Lewis Steele, Jr., and Gerald Jay Sussmann, is my first love among programming languages, though I haven’t always been faithful. There’s a new draft standard for Scheme now circulating at http://www.r6rs.org/. The proposal is open for comments until next March; on final approval it will become “The Revised⁶ Report on the Algorithmic Language Scheme.”

Scheme has always been a language under both tension and pressure. It started out (more than 30 years ago) as a sandbox for playing with some new ideas in programming-language semantics, and also (I think) as an attempt to “do Lisp right.” Soon it became an important teaching language, and it still is. But there’s always been pressure to make more of it. Shouldn’t a language for grownups have modules, libraries, foreign-function interfaces—not to mention classes, methods, inheritance and all the other apparatus of OO? That’s the pressure part. The tension has come from unceasing debates over issues such as the ideal syntax and semantics of macros, and the morality of creating a version of Lisp without an eval procedure.

I’ve only just begun to read the new draft, which is officially the Revised^5.91 Report. But I think one can draw some preliminary conclusions just from its heft. Here are the page counts of the principal versions of the standard:

An introductory essay that first appeared in the Revised³ Report begins:

Programming languages should be designed not by piling feature on top of feature, but by removing the weaknesses and restrictions that make additional features appear necessary.

The same text is still present in the new document. I’ll be interested to learn whether the same spirit has survived. It’s not to be taken for granted that just because the standard has doubled and then redoubled in size, the language has also grown obese. But it’s an obvious worry.